I recently had an intense debate about music, art, and the role of generative AI in creativity with an old friend and fellow musician. He has been using the platform Suno to… create?
The things you can “create” on Suno are stunning. Not good, but definitely stunning. It’s forced me to reflect deeply on what I care about in music and in the act of creation.
For a long time I would have said that the point of music – of all art, really – is to evoke emotion, and I think that’s a fairly widespread theory of art. But long before gAI came along, that theory no longer satisfied me. Once you know enough about music theory and audio production, you know that we’re pretty much programmed to react emotionally to certain rhythmic, harmonic, and sonic patterns.
Four-on-the-floor at 120bpm makes you want to put your hands in the air and wave them like you just don’t care. Playing the root’s major third over the IVth chord makes you yearn like your first crush. A chorus of scooped overdriven guitars makes you ready to go out and bust some heads.
It’s amazing that these devices work, and it’s an admirable craft to learn how to deploy them.
Seeing through the devices doesn’t mean music no longer achieves its emotional impact. But it’s made me recognize something I’ve known intuitively all along: What I value in my favorite works is not how they shape my inner life. It’s the insight they give me into someone else’s.
Escaping solipsism
Handprints in caves from tens of thousands of years ago aren’t particularly beautiful or moving. But to this day, they inspire awe because someone thought descending into a dark cave and leaving their handprint was a cool thing to do. Someone who was clearly like us.
We experience the world through sense impressions, and then our minds construct a reality out of them. We update that interior simulation constantly to better match the sense impressions the exterior reality bombards us with.
That’s true of the physical world of three dimensions, and equally true of the social world of strategic interactions. We run little simulations of the people we interact with. Ideally, we update those models to ever more closely match the real, live people we love, collaborate, and negotiate with. But often we refuse to update those models, reducing people to stereotypes to suit our purposes. Or just to keep our heads above the flood of data.
Most of us do this completely unconsciously all the time. Some of us are aware of what’s going on some of the time. Either way, though, we run the risk of losing ourselves in our toy world populated by toy people. The downside is not just that our map might drift from the terrain. The downside is also the supreme loneliness of solipsism.
The music, the writing, the art that I have always loved the most is not necessarily what I’ve found aesthetically pleasing or deeply moving. It’s the art where I’ve most clearly seen the mind on the other side, delighting in the process of creation as I delight in it. To me, songs like Bohemian Rhapsody and Wuthering Heights are like coded signals coming through the prison walls of our own minds.
I find it hard to describe what it is exactly that makes something a secret signal and not just someone “strumming my pain with his fingers, singing my life with his words.” You know it when you hear it: someone, a real person, not my internal simulation of one, was experiencing joy in the creative act.
Next to being present with the people closest to us, appreciating art is one of the most powerful ways to encounter an autonomous mind. In one way at least, it’s more powerful because it transcends time and space.
Blocking the exit
Today, a song generated by Suno with nothing more than a 20-word prompt can be quite moving, at least as moving as what’s been filling the commercial radio airwaves for the past couple of decades.
But if I know something was created by AI, in what sense could I infer the existence of another mind delighting in the creative act?
Conversely, if I know a lot of art is AI-generated, how should I think about the secret signals from other minds I believe I hear? The mere possibility I might be fooled makes all signals suspect.
Generative AI may produce aesthetic pleasure and emotional impact. It’s not there yet, but it may do so more efficiently, at lower cost, and maybe more reliably and powerfully if the algorithms more perfectly push our buttons.
But is that worth the cost of losing one of the best ways to see we’re not the only autonomous beings in this big, beautiful, but sometimes very lonely universe?
The Ruminathans 